! "#$ "%&'" (' ! ) #&
*+ & , & -. / . , &0 , . &( 1, 23)!&4 . . . & & 5 6 & , "#7 #' 6 . & #' 6 & . , & . .3( & ( & 8 !10 92&4 . ) *)+ . : ./ *)#+ . *) +& . . , &
"#$ :;; &&;< = : ::: #>$ ?
4@!%$5%#$?>%'>'> 4@!%$5%#$?>%'>>#
!
#"?%#
TOP QUARK AND HEAVY VECTOR BOSON ASSOCIATED PRODUCTION AT THE ATLAS EXPERIMENT
Olga Bessidskaia Bylund
Top quark and heavy vector boson associated production at the ATLAS experiment
Modelling, measurements and effective field theory
Olga Bessidskaia Bylund ©Olga Bessidskaia Bylund, Stockholm University 2017
ISBN print 978-91-7649-343-4 ISBN PDF 978-91-7649-344-1
Printed in Sweden by Universitetsservice US-AB, Stockholm 2017 Distributor: Department of Physics, Stockholm University Cover image by Henrik Åkerstedt Contents
1 Preface 5 1.1 Introduction ...... 5 1.2Structureofthethesis...... 5 1.3 Contributions of the author ...... 6
2 Theory 8 2.1TheStandardModelandQuantumFieldTheory...... 8 2.1.1 Particle content of the Standard Model ...... 8 2.1.2 S-matrix expansion ...... 11 2.1.3 Cross sections ...... 13 2.1.4 Renormalisability and gauge invariance ...... 15 2.1.5 Structure of the Lagrangian ...... 16 2.1.6 QED ...... 17 2.1.7 QCD ...... 18 2.1.8 Weak interactions ...... 20 2.1.9 Electroweak interactions ...... 22 2.1.10 Electroweak symmetry breaking ...... 22 2.2TestsoftheStandardModel...... 23 2.3 Questions not answered by the Standard Model ...... 24 2.4 Motivation for ttV¯ searches...... 25 2.5 The treatment of the Wt and tW Z processes with Diagram Removal...... 26 2.6EffectiveFieldTheory...... 28
3 Experimental setup 32 3.1 The Large Hadron Collider ...... 32 3.2 The ATLAS experiment ...... 33 3.2.1 Inner Detector ...... 34 3.2.2 Liquid Argon Calorimeter ...... 36 3.2.3 The Tile Calorimeter ...... 37 3.2.4 Muon system ...... 39 3.2.5 Magnet system ...... 41 3.2.6 Trigger system ...... 41
4 Measurements of the ttV¯ production cross sections 43 4.1 Object definitions ...... 44 4.1.1 Jets ...... 44 4.1.2 Electrons ...... 45 4.1.3 Muons ...... 46 4.1.4 B-jets ...... 46 4.1.5 Missing ET ...... 47 4.1.6 Overlap removal ...... 47
1 4.2 Signal regions ...... 47 4.3 Control regions ...... 51 4.4 Fake leptons, overview ...... 51 4.5 Background estimation with Diagram Removal ...... 53 −1 4.6 Analysis I at 8 TeV, 20.3fb ...... 58 4.6.1 Event yields I ...... 58 4.6.2 The WZ background I ...... 59 4.6.3 Results I ...... 59 −1 4.7 Analysis II at 13 TeV, 3.2fb ...... 62 4.7.1 Yields II ...... 62 4.7.2 WZ II ...... 62 4.7.3 Results II ...... 62 4.8 Analysis III at 13 TeV, 36 fb−1 ...... 66 4.8.1 Yields III ...... 66 4.8.2 Redefinition of trilepton regions for analysis III .... 68 4.8.3 The WZ background III ...... 71 4.8.4 Fake leptons III ...... 72 4.8.5 Results III ...... 75
5 Effective field theory 77 5.1 Cross section and distributions with effective operators .... 77 5.2 EFT limits, simultaneous fit ...... 85 5.3 EFT simulations for ATLAS ...... 88 5.4 EFT limits from the 13 TeV measurement ...... 94
6 Conclusions and outlook 96 6.1 Uncorrelated systematic uncertainties ...... 107 6.2 Correlated systematic uncertainties ...... 107
7 Sammanfattning 109
2 Acknowledgements
I would like to thank my supervisor Jörgen Sjölin for helping me choose the subject and to find my niches in the field, which have been more to the phenomenological side. I am grateful to my co-supervisor Sten Hellman and my team leader David Milstead for their feedback and support in the writing of this thesis. I am happy to be part of a supportive and positive group in Stockholm. Katarina in particular has been a rock. Many thanks to the colleagues and friends who helped proof read sections of my thesis. All the publications that this thesis is based on are the result of collab- orative work. In this thesis I highlight my contributions while also giving the overall context and results. Below, I acknowledge the contribution of my collaborators in the cases where we were working on closely related subjects. Hovhannes Khandanyan (Stockholm) developed the software and defined the signal regions of the trilepton channel that were used for the ttV¯ analysis I. I evaluated the systematic uncertainty on the WZ background for analysis I together with Kerim Suruliz (Sussex) and Jörgen Sjölin. Kerim provided the simulations in Sherpa, which I used to compute transfer factors and the largest extrapolation uncertainties for the WZ process for analysis I. For the ttV¯ analyses II and III, the WZ extrapolation uncertainties were evaluated by Kerim; I cite these results for completeness. The fake rates for analysis II and III were derived by Jörgen Sjölin. I was involved in the validation of the fake rates for analysis III. The final results for the ttV¯ analyses that are shown in this thesis were obtained by Tamara Vasquez Schroeder in analysis I and by Maria Moreno Llacer (CERN) for analysis II. Maria also redefined the regions in the same sign dilepton channel for analysis III, which gave us a higher expected sen- sitivity for the ttW¯ measurement. The fake factors used to scale the γ +X background for analysis III were derived by Alexandra Schulte (Mainz). I would like to thank Kerim for demonstrating the functionality of the software for analysis III, which allowed me to design the fit for the EFT coefficients in a nice way. Many thanks to Karl Gellerstedt (Stockholm) for dressing the ttl¯ +l− EFT samples so that I could derive the parameters for the EFT fit for analysis III in regions that closely resemble the signal regions of our analysis. I am grateful to Fabio Maltoni (UCLouvain) who welcomed me to his team and to the MCnet studentship program, which allowed me to spend three and a half months in Belgium, learning about Effective Field Theory and the Diagram Removal methods. For the EFT phenomenological analysis that I performed when I was based in Louvain, I worked together with Ioan- nis Tsinikos on simulating the ttμ¯ +μ− process. We discussed the strategy together with my supervisor for the time Fabio Maltoni and collaborators Eleni Vryonidou and Cen Zhang. Cen had provided the code describing
3 the top quark interactions in the framework of Effective Field Theory and patiently answered my questions that arose while I was working on he EFT fit for analysis III. I learned the Diagram Removal 2 method from Federico Demartin (UCLouvain) and had many discussions with him. On a more personal level, I would like to thank Ioannis for being my gym buddy and Joze for cooking for me while I was in Belgium, you made my stay more enjoyable. Thank you to my figure skating coach Susanne and skating friends Natasja and Camilla for the mental training and a great summer. Thank you to Henrik Åkerstedt for the photos with the ATLAS detectors and for many memorable adventures. I am very grateful for my big loving family and happy to have great friends. Thank you for helping me through the tougher periods and for lifting me up higher in good times. An extra high five goes to Ea, Mike, Lotta, Mariana, Surre, Tomas K. and Stefan. To my husband Tomas, you have been incredible.
4 1 Preface
1.1 Introduction The Standard Model (SM) of particle physics [1], [2] describes our best understanding of particle physics at present. The predictions of the Standard Model have been confirmed by the discoveries of the particles it predicts and by measuring their interactions. Some notable examples of this are the confirmation of electroweak unification by the discoveries of the W and Z bosons 1982 and 1983 [3] and the discovery of the sixth and heaviest quark, namely the top quark, by the CDF and D0 experiments at the Tevatron in 1995 [4]. In order to produce the heaviest particles of the SM and to study their properties, large energies are required. For this purpose, proton beams are accelerated to close to the speed of light inside the Large Hadron Collider (LHC) [5] and brought to collide at the centre of the ATLAS experiment [6]. The protons scatter against each other, creating new particles, which are registered in the ATLAS detector and then the collision events are analysed. In this thesis, the physics of top quarks is explored. The LHC acts as a top quark factory, producing millions of top quark pairs each year. The main focus of this thesis is on the production of top quark pairs in association with a vector boson (Z or W ) at the ATLAS experiment. In order to model an important background to ttZ¯ , namely single top associated tW Z production, at next-to-leading order in QCD, the resulting overlap with other processes needs to be removed. Two varieties of the Diagram Removal method are used for this purpose. The Standard Model hypothesis for ttZ¯ and ttW¯ production is tested against data. The scenario where ttZ¯ production is affected by the presence of new physics is also explored, using the Effective Field Theory (EFT) approach [7], [8], [9]. Effective Field Theory is a model-independent framework that parametrises the new physics in terms of different operators that give rise to various of interactions of the SM particles. No particles beyond the SM are assumed to be produced at the energies accessible at the LHC. Instead, the interactions of known particles are assumed to be modified by the effective operators. The coupling constants for the effective operators are constrained using the expected sensitivity of a measurement of ttZ¯ production performed by ATLAS.
1.2 Structure of the thesis This thesis is structured as follows. Section 2 gives a theoretical background for the work presented in this thesis, covering the Standard Model, the Dia- gram Removal methods and Effective Field Theory. In Sec. 3, an overview of the experimental apparatus is given, with the Large Hadron Collider and the
5 ATLAS experiment described. It should be noted that this section overlaps considerably with the licentiate thesis in Ref. [10]. Three measurements of ttZ¯ and ttW¯ production at ATLAS are presented in Sec 4. The first measurement [11], referred to as analysis I, was performed on data from collisions with a center of mass energy of 8 TeV. In analysis II the measurement was carried out at a higher collision energy of 13 TeV, at −1 an integrated luminosity of 3.2fb [12]. The measurement of ttZ¯ and ttW¯ production was then extended for analysis III, adding the 2016 dataset to give a total of 36 fb−1. This analysis is not yet public, at the time of writing, instead the currently expected results are quoted. At the start of Sec. 4, an overview is given of the strategy and the physics objects that the three analyses have in common. Here Sec. 4.1 intersects with the licentiate thesis, Ref. [10]. After a common description, the specific methods and results for each analysis are presented in Secs. 4.6-4.8. In the modelling of the ttZ¯ signal, the Z boson is required to decay into two charged leptons. The contribution and interference of an off-shell photon ttγ¯ ∗ giving rise to the same final state is taken into account by specifying ttl¯ +l− in the simulation (without specifying requirements on the intermediate boson). In the last chapter, Sec. 5, distributions in ttl¯ +l− in the presence of effective operators are shown. Fits for the coefficients are performed to data. First a toy global fit is done, using several measurements at once to fit several coefficients, while making broad assumptions about the correlations between the uncertainties, as is described in Sec. 5.2. Next, in Sec. 5.4, fits to one coefficient at a time are performed, using the framework for analysis III.
1.3 Contributions of the author My contributions to the different projects are summarised below. For analysis I, my focus was on trilepton channel, targeting ttZ¯ and ttW¯ production; this is the most sensitive channel for ttZ¯ . I studied the fake lep- ton background and the WZ background for this channel and maintained and synchronised our software to the other groups within ATLAS and, at a later stage, with CMS. The treatment of fake leptons is described in detail in my licentiate thesis [10] and only mentioned briefly here. For WZ pro- duction, the central value is measured in a dedicated control region. The uncertainty from the measurement is referred to as the normalisation un- certainty. Additionally, extrapolating the WZ production from the control region into the signal regions is associated with systematic uncertainties. I evaluated the normalisation uncertainty and used one of the two alternative methods to evaluate the extrapolation uncertainties for analysis I. The central value and normalisation uncertainty for the WZ production cross section were determined with the same method in analyses II and III, with some minor modifications to the definition of the relevant control
6 region. Additionally, for both of these analyses, I provided the prediction and modelling uncertainty for the tW Z background. I simulated this process at next-to-leading order in QCD using the Diagram Removal 1 method and assessed the uncertainty by comparing inclusively with the Diagram Removal 2 prediction. Using the DR1 prediction resulted in a 20 % lower predicted yield for the background process tW Z than one would obtain at LO (as was done in analysis I. By studying the DR1 and DR2 predictions for tW Z it became apparent that this process was not well understood, which served as an argument against lowering the assigned systematic uncertainty on tW Z for analyses II and III. The Diagram Removal methods applied to tW Z as well as Wt were presented in Ref. [13] and summarised in the conference proceedings [14] for the International Top Workshop 2016. For the work described in Ref. [13], I was responsible for the chapter on Diagram Removal. I also generated DR1 and DR2 simulations of Wt production for ATLAS in MG5_aMC@NLO. For analysis III, I worked on the same sign and trilepton channels and for six months of the analysis, I was one of the two main editors of the in- ternal documentation and combined the results from the different channels. Moreover, I provided ttl¯ +l− predictions, modified by five different effective operators and performed a fit for the coefficients. The effects of the op- erators on various top quark physics processes was presented in Ref. [15]. My contribution to this paper was discussing the strategy, giving the exper- imental perspective, simulating the ttl¯ +l− process and looking into many distributions for this process. I also performed a fit, constraining all rele- vant coefficients at the same time (using some simplified assumptions) to measurements performed at the LHC (by 2015), but this was not included in the paper; this work is described in Sec. 5.2. In addition to this, as part my PhD programme, I studied the electronic noise in the Tile Calorimeter. This work is described in Ref. [10] and is not repeated here. I also spent one month in 2013 replacing power supplies and doing consolidation work for the Tile Calorimeter.
7 2 Theory
In this chapter, the Standard Model (SM) is described in Sec. 2.1. Successful tests of the SM are mentioned in Sec. 2.2. Problems with the Standard Model are highlighted in Sec. 2.3 and the measurements of ttZ¯ and ttW¯ production are motivated in Sec. 2.4. The Diagram Removal methods are introduced in Sec. 2.5 and in Sec. 2.6, the effective operators of interest for ttl¯ +l− production are shown.
2.1 The Standard Model and Quantum Field Theory The Standard Model of particle physics [1], [2] describes the particle content of the universe and the interactions of the particles. It is a theory of quantum mechanical fields φi(x), with particles described as the quanta of the fields. The energy density of the system is given by the Lagrangian density L(x), which is a function of different fields φi(x) and their derivatives ∂μφi(x). Integrating the Lagrangian density over the space coordinates d3x gives the Lagrangian. The Lagrangian density consists of a free field part L0(x), which includes a kinetic and a potential term, and an interaction part LI (x), which describes interactions of different kinds of fields. In this thesis, charges are given in units of the elementary charge (the magnitude of the charge of the electron) and natural units are employed, where c = = 1. In natural units, the dimensions of mass m and energy E are the same and equal to the dimension of length L inverted [m]= [E]=[L]−1. The action, from which the interactions of the particles can be derived, is given by: